skill-creator
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute local Python scripts included in the package (
scripts/init_skill.pyandscripts/package_skill.py) to automate directory creation, file generation, and zip packaging for new skills. - [PRIVILEGE_ESCALATION]: The
init_skill.pyscript applies executable permissions (chmod 0o755) to a dynamically created example script (scripts/example.py). This is a functional requirement for the skill's purpose as a project initializer. - [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface where it ingests external data from a URL and utilizes file-writing capabilities.
- Ingestion points: Fetches content from
https://code.claude.com/docs/en/skillsusingWebFetchinSKILL.md. - Boundary markers: The prompt uses specific instructions to extract structured information, but lacks explicit ignore-markers for embedded instructions.
- Capability inventory: Includes file system writes, directory creation, and permission modification via
scripts/init_skill.py. - Sanitization: None detected for the fetched web content before it is processed by the agent.
- [EXTERNAL_DOWNLOADS]: Fetches documentation and guidelines from
code.claude.comto align skill creation with official standards.
Audit Metadata